Cross-genera SSR transferability in cacti revealed by a case study using Cereus (Cereeae, Cactaceae)

Abstract The study of transferability of simple sequence repeats (SSR) among closely related species is a well-known strategy in population genetics, however transferability among distinct genera is less common. We tested cross-genera SSR amplification in the family Cactaceae using a total of 20 heterologous primers previously developed for the genera Ariocarpus, Echinocactus, Polaskia and Pilosocereus, in four taxa of the genus Cereus: C. fernambucensis subsp. fernambucensis, C. fernambucensis subsp. sericifer, C. jamacaru and C. insularis. Nine microsatellite loci were amplified in Cereus resulting in 35.2% of success in transferability, which is higher than the average rate of 10% reported in the literature for cross-genera transferability in eudicots. The genetic variation in the transferred markers was sufficient to perform standard clustering analysis, indicating each population as a cohesive genetic cluster. Overall, the amount of genetic variation found indicates that the transferred SSR markers might be useful in large-scale population studies within the genus Cereus.

Simple sequence repeats (SSR) or microsatellites are, in general, non-coding regions commonly found in Eukaryote genomes composed of tandemly arranged repeat motifs from 1 to 6 base pairs (Oliveira et al., 2006).SSRs are useful molecular markers for several applications in population genetics and breeding studies, as they frequently exhibit high levels of polymorphism, in addition to their abundance and random distribution across and throughout genomes.In plants, SSR loci have been used for several purposes, for example, estimates of genetic diversity (Zhu et al., 2016), intra-and interspecific gene flow (Palma-Silva et al., 2011;Pinheiro et al., 2014), biogeographical distributions (Beatty and Provan 2011), phylogenetic relationships (Mehmood et al., 2016), genetic mapping (Tan et al., 2016), and conservation (Gómez-Fernández et al., 2016).
An alternative to overcome time consuming and costly development of a new set of SSR primers for a target species is to carry out the transferability of SSR primers among related species (Barbará et al., 2007;Lavor et al., 2013;Nogueira et al., 2015).The rate of success in this approach (i.e., heterologous amplification) depends on the nucleotide similarity among the flanking regions of different species.Therefore, it is expected that there will be a higher rate in heterologous amplification among taxa with recent divergence times.In plants, this technique has been widely adopted for a great variety of eudicots (e.g., Haerinasab et al., 2016;Mengistu et al., 2016), where the average rate of success at infrageneric level is around 60% (Barbará et al., 2007).The rate of cross-genera transferability is around 10% in eudicots (Barbará et al., 2007), but the levels of success may reach values above 50% in some plants (Satya et al., 2016).
Taking into account the recent divergence within Cactaceae, as well as its emergence as an informative model to study diversification in xeric habitats (Arakaki et al., 2011), the aim of this study was to perform cross-genera SSR amplification in four closely related taxa of the genus Cereus (Cactaceae; Cereeae) occurring in eastern Brazil: C. fernambucensis subsp.fernambucensis, C. fernambucensis subsp.sericifer, C. insularis and C. jamacaru.A previous phylogenetic analysis based on plastid DNA placed C. jamacaru as a member of a polytomic clade, sister of the monophyletic clade composed by C. fernambucensis and C. insularis (Franco et al., 2017a).In this study, we selected a set of 11 SSR loci originally described for Ariocarpus bravoanus (Hughes et al., 2008), Echinocactus grusonii (Hardesty et al., 2008) and Polaskia chichipe (Otero-Arnaiz et al., 2004) that were recently transferred to Cereus species cultivated in different urban areas, including C. hildmannianus (Martin 2011;Fernandes et al., 2016).An additional nine SSR loci described for Pilosocereus machrisii (Perez et al., 2011) were included in this investigation.
We sampled 122 individuals from representative populations of C. jamacaru, C. insularis and C. fernambucensis (Table 1), besides one individual of C. hildmannianus (Salto, SP;23.99, 47.33;SORO 2746) as a positive control in the initial tests.Genomic DNA was extracted from the radicular tissue using the Qiagen DNeasy Plant Mini Kit (Qiagen, Hilden, Germany).As the samples from localities S82 and S83 are geographically (~34 km) and genetically close, sharing the same unique alleles and comprising a cohesive genetic group, we decided to join the individuals from the two populations in a single sample, hereinafter referred to as S82/S83.Initial amplification tests were performed using a subsample of 12 individuals, with slight modifications on PCR conditions as described by Albert and Schmitz (2002), Don et al. (1991), andPerez et al. (2011).The reactions were performed in 10 mL of total PCR volume including 0.5 U of Taq DNA Polymerase (Promega), 1X Taq Buffer (5X Colorless GoTaq® Flexi Buffer), 0.2 mM dNTPs, and primer and MgCl 2 concentrations varying when necessary.We considered a locus successfully transferred when the PCR products were clearly visualized in 3% agarose gels and showed a product size compatible with the range described for that locus.The loci successfully amplified were then genotyped in the total sample (Table 1) using PAGE (denaturing polyacrylamide gel) with concentrations varying between 6% to 9%, according to expected allele size.To visualize the alleles, the gels were stained with silver nitrate.The percentage of transferability success was estimated according to the number of individuals amplified in each locus.
From the 20 tested loci (Table S1), nine (mAbR 28 from A. bravoanus; mEgR 02, mEgR 76 and mEgR 78 from E. grusonii and Pmac82, Pmac84, Pmac108, Pmac146 and Pmac149 from P. machrisii) showed positive results in transferability for at least one species (Table S2), resulting in 35.16% of success in transferability.Except for mEgR 02, the allele size for all loci was congruent with the expected size (Table S2).We were not able to amplify five SSR loci previously transferred to Cereus (Pchi 21, Pchi 47, Pchi 54, mAbR 42 and mAbR 77) (Martin, 2011;Fernandes et al., 2016), even after several attempts to modify PCR conditions (Table S3).This result is likely related to nucleotide differences among the flanking regions of the samples used in this work, preventing primer annealing.
The percentage of polymorphic loci ranged from 44.4% (populations S113 and S82/S83) to 77.8% (location S80) (Table 2).In contrast with the expectation of reduced levels of genetic diversity at transferred SSR loci (e.g.Goldstein and Pollock, 1997;Jan et al., 2012;Moodley et al., 2015), we found higher levels of polymorphism for some loci (Pmac82 in all populations, mEgR 78 in S88 and mEgR 02 in S80) than those reported in the original description (Table 2).No locus showed significant heterozygosity deficiency in relation to expectations of HWE after Bonferroni correction.Inbreeding coefficient estimates (F IS ) provided no significant result (Table 2).The locus Pmac108 showed high levels of observed heterozygosity in all populations, excepting S82/S83 ( leles were found in populations S113, S88, S115D, S114, S104 and S82/83 (Table S4).The LD analysis results between polymorphic loci were not statistically significant after Bonferroni correction.
The FreeNA-corrected estimate of global F ST was 0.44, ranging from 0.12 (Pmac82) to 0.79 (mEgR 02) (Table S5).Clustering analyses (NJ, PCoA, STRUCTURE) have shown somewhat distinct results (Figure 1 and Figure S1).However, the results from AMOVA suggest that the three clusters recovered by STRUCTURE better explain the genetic variation structuring in our data (Table 3), as follows: 1) S113, S014 and S115D populations; 2) S80, S88 and S114 populations; 3) S82/S83 (Figure 1a).To investigate sub-structuration within our data, we performed a STRUCTURE analysis for each cluster, which resulted in each location being a cohesive genetic group (Figure 1b).Although we have not done an extensive geographic sampling for each studied taxon, and the number of markers is relatively low, some clustering results recovered here agree with previous phylogeographic hypotheses established for C. fernambucensis and C. insularis based on cpDNA and the PHYC gene (Franco et al., 2017b)   F SC = 0.09764Among populations within groups 6.37 but seems to be a spurious grouping as a result of the reduced number of sampled populations.The estimated success in transferability observed in this study (35.16%) was higher than the average of 10% found in cross-genera transferability studies published between 1997 and mid-2006 (see Barbará et al., 2007).However, this is not an uncommon result, as similar findings or even higher levels of cross-genera transferability were observed in some groups of plants (Satya et al., 2016).In the family Iridaceae, for example, a success of 77% was observed in cross-amplification between genera (Miz et al., 2016).In the family Malvaceae cross-genera SSR transferability varied from 71% to 92% (Satya et al., 2016), while in Euphorbiaceae these percentages ranged from 9.5% to 59.1% (Whankaew et al., 2011).Evidently, genera are taxonomic categories mainly based on morphological instead of genetic information, and different levels of phylogenetic divergence must be embedded within each genus.Therefore, the success in cross-genera transferability may vary highly depending on the target organism.On the other hand, it is expected that the success in cross-amplification should be a function of phylogenetic distance, at least regarding genetic differentiation (Barbará et al., 2007).This expectation was not clearly observed here considering cactus phylogeny (Hernández-Hernández et al., 2014).We observed similar success in heterologous amplification using primers described for relatively distantly (A.bravoanus and E. grusonii -four transferred of nine tested) or closely related species (P.machrisii -five transferred of 11 tested).It is worth highlighting that for those loci from A. bravoanus, E. grusonii, and P. chichipe we had a previous expectation of positive cross-amplification, as they were formerly transferred for some Cereus species (Martin, 2011;Fernandes et al., 2016).Nevertheless, higher levels of cross-genera amplification in the cactus family might be a widespread tendency, as the main lineage divergences and species radiation events within this family are thought to have occurred in the last 10 Myr (Arakaki et al., 2011;Hernández-Hernández et al., 2014;Silva et al., 2018).Even with remarkable morphological distinctness among cactus species, resulting in more than 120 recognized genera (Hunt et al., 2006), these recent divergence times increase the possibility of heterologous amplification in the family Cactaceae due to the expected similarities in the flanking SSR regions among different species.Evidently, this is not a rule, as even here we found discordance between some results obtained by Martin (2011) andFernandes et al. (2016), which were likely due to nucleotide differences in flanking regions of the distinct samples.At any rate, this information should be taken into consideration to encourage cross-genera transferability studies in Cactaceae, which, despite their potential, are still relatively scarce in this family (Table S6).
The genus Cereus constitutes an interesting biological model to perform evolutionary studies, and efforts were employed to screen informative molecular markers in this genus (Silva et al., 2016;Romeiro-Brito et al., 2016) to solve species level phylogeny (Franco et al., 2017a) and to investigate population differentiation and phylogeography (Franco et al., 2017b;Silva et al., 2018).Our results are in line with these endeavors, supplying additional molecular markers that can be useful for estimating genetic diversity and gene flow in target Cereus species.Furthermore, considering the rate of success in transferability, our results should encourage cactus researchers interested in using the increasing number of SSR loci that have been described for representatives of this highly diverse and relatively overlooked plant family (e.g., Bonatelli et al., 2015;Fava et al., 2016).

Figure 1 -
Figure 1 -Population differentiation in STRUCTURE, (a) results for K = 3 on the first level structure, and (b) separating each population as a distinct genetic group.The southern and northern population groups of C. fernambucensis subsp.sericifer and C. fernambucensis subsp.fernambucensis are based on phylogeographic circumscription(Franco et al., 2017b).

Table 1 -
Geographical localities of the populations from three species of Cereus used in this work.

Table 2 -
Genetic diversity indices: Number of samples (N), Number of alleles (Na), Effective allele number (Ne), Observed heterozygosity (Ho), Expected heterozygosity (He), absence (-) and presence (+) of null alleles, and F IS values per loci and population are shown.
. The close relationship of C. jamacaru and C. fernambucensis subsp.

Table 3 -
Global and hierarchical Analysis of Molecular Variance (AMOVA).For hierarchical AMOVA the a priori groups are based on taxonomic circumscription, NJ phenogram, PCoA and STRUCTURE.